Nonlinear characteristics of observed solar radiation data

The solar radiation data is one of the most important indicators for photovoltaic (PV) power generation application. It is utilized frequently for system capacity plan, operation and dispatch, reliability evaluation, system modeling and simulation of photovoltaic power station. Faced with the nonlinear, non-stationary characteristics of the observed solar radiation data, a detailed study is achieved in this investigation. Some observed data from North American solar radiation stations is implemented to analyze their box dimensions and Hurst exponent indexes firstly, which are the single scale fractal characteristics. Then, the Multi-fractal Detrended Fluctuation Analysis (MDFA) method is employed to study its multi-fractal characteristics. In addition, the power-law distribution is found in statistics and power spectrum scales, which can describe its heavy tail, nonlinear, non-stationary characteristics. Results show that the box dimensions and Hurst exponent indexes in daily or yearly scale are close to constants and vary in the location of stations; the MDFA results show its obvious multi-fractal characteristics, which are affected by time and location factors; the power-law distribution of solar radiation data in statistics and spectrum scales reveals its non-stationary characteristic, and the power-law index is in a narrow range. These results have significant value and reference for the PV power plant location selection, capacity determination, solar radiation prediction, and operation schedule. It has explored a new way for solar radiation data research as well.

► We have found the nonlinear characteristics of solar radiation.
► Box dimension and Hurst index are utilized to demonstrate its single scale fractal.
► Multi-fractal detrended fluctuation analysis is employed for its multi-fractal.
► Its power-law distribution is explored in the statistical characteristics.
► Power laws of the solar radiation data in power spectrum are found.